Checks

A check is a scheduled probe against an endpoint. WatchDeck currently supports two types — pick the one that matches what you’re monitoring.

Adding your first check

Adding an endpoint is adding a check — the form lives at Endpoints → Add endpoint. It’s a single page with four labelled sections (General, Monitoring, Assertions, Alerts), not a wizard, so you can jump between sections without losing state.

How a check runs

Each scheduled run flows through the same pipeline regardless of type:

1. Probe. WatchDeck issues the request — an HTTP fetch, or a TCP connection attempt.
2. Status evaluator. Decides healthy / degraded / down from the raw probe result. Rules in order: connection error → down; port refused → down; HTTP status not in expectedStatusCodes → down; response time over latencyThreshold → degraded; otherwise healthy.
3. Assertions. If the base status isn’t already down, every configured assertion runs against the response. The worst failed severity (down or degraded) wins.
4. Streak counters. A run that’s down increments the failure streak. Once the streak reaches failureThreshold, an incident opens. A run of recoveryThreshold consecutive healthy results closes it.

Defaults

Editing a check after creation

Open the endpoint and switch to the Settings tab. Every field from the add form is editable live, with one exception: the check type (HTTP vs TCP) is locked at creation. To change type, use Clone Endpoint from the endpoints list and pick the new type when the form opens.

Previewing before saving

The Assertions section has a Test now button that runs a one-shot probe against the configured URL and renders the live response — status code, latency, content type, and per-assertion pass/fail — without writing anything to your check history. Use it to sanity-check assertions while you’re still drafting them.

See Assertions → Testing live for the full preview flow.

What’s next